Next-generation multiplex-edited CAR-NK cells: more edits, more power?

Abstract

First clinical trials demonstrated the safety of adoptive cell transfer with allogeneic natural killer (NK) cell products from healthy donors, making them an attractive candidate for ‘off-the-shelf’ chimeric antigen receptor (CAR)-immune cell therapy. However, reduced persistence and inactivation of NK cells by immunosuppressive cues likely limit the performance of CAR-redirected NK cells. Wang and colleagues demonstrate that multiplex CRISPR base editing allows optimization of the intrinsic functionality of CAR-NK cells improving their therapeutic potential. In contrast to conventional CRISPR-Cas nucleases, base editing avoided most double-stranded DNA breaks while enabling highly efficient editing at up to six sites simultaneously. The study further demonstrates the feasibility of a non-viral approach to integrate CAR transgene and multiplex base editing of several immune checkpoints in NK cells using a single electroporation. CAR-NK cells harboring up to three base edits demonstrate improved potency over unedited counterparts in vitro. Xenograft mouse models confirmed increased potency, but also indicated signs of organ toxicity - a phenomenon that will require future studies prior to clinical translation. The study demonstrates that CRISPR base editing is a powerful tool to unleash the full cytotoxic potential of NK cells, but it also warrants the question: How many internal breaks can be removed without hurting CAR-NK cell therapy’s impeccable safety record?

Commentary

Natural killer (NK) cells are becoming an increasingly important modality for cellular immunotherapy of cancer.¹ The intrinsic cytotoxic capacity and their ability to recognize infected and transformed cell lacking major histocompatibility complexes make them a promising approach in targeted cancer therapy.² They provide a versatile tool by complementing pathogen recognition with their intrinsic antitumor killing capacity induced by a range of natural cytotoxicity receptors and their ability to destroy antibody-bound cells via antigen-dependent cellular cytotoxicity (ADCC).² Additionally, the absence of rearranged antigen receptors allows for their use in allogeneic, ‘off-the-shelf’ therapies. Early clinical trials using allogeneically administered primary NK cells have demonstrated significantly improved safety profiles compared with chimeric antigen receptor (CAR) T cell therapies, which can be associated with risks such as graft-versus-host disease, cytokine release syndrome, and neurotoxicity.³

Genetic modification is a powerful tool to enhance the therapeutic efficacy of NK cell therapy. The integration of CAR constructs into NK cells can enhance their tumor-targeting capabilities and CAR-NK cells show promising clinical responses in early phase clinical trials.³ Co-delivery of NK-cell supporting cytokine transgenes, such as interleukin (IL)-15, promotes optimal activation and superior CAR-NK cell function in vivo.³ Non-viral techniques, such as transposons or CRISPR-Cas mediated knock-in, offer the potential for more cost-effective clinical translation over viral vectors.4,6 However, early clinical data indicate that although CAR-NK cells show favorable safety profiles, the tumor response rates and limited NK cell persistence still leave room for optimization.

The study by Wang et al¹ published in the current issue, directly addresses these challenges by developing a fully non-viral platform for the precise engineering and functional enhancement of CAR-NK cells using CRISPR-Cas base editing (BE) (figure 1). The authors developed a highly efficient protocol for gene editing of primary NK cells using an adenine base editor (ABE) composed of catalytically inactive Cas9 fused to a deaminase domain for site-specific nucleotide conversion. This approach allows the introduction of gain- or loss-of-function mutations without inducing double-strand breaks (DSBs) or requiring a donor DNA template. BE provides a clear advantage over conventional CRISPR-Cas9 for multiplex gene editing, which can cause translocations, chromosomal rearrangements and toxicity by inducing multiple DSBs simultaneously.

Figure 1. Generation and characterization of multiplex base-edited CAR-IL15 NK cells. Schematic illustration: Natural killer (NK) cells were isolated from allogeneic healthy donors. A single-step, virus-free electroporation approach was employed to achieve simultaneous TcBuster transposon-mediated integration of a CD19-specific chimeric antigen receptor (CAR) and interleukin-15 (IL-15), combined with multiplex base editing enabling up to six precise genomic modifications. The resulting engineered NK cells (multiplex base-edited CAR-IL-15 NK cells) exhibited reduced double-strand DNA breaks and chromosomal translocations. They demonstrated enhanced antilymphoma activity and prolonged in vivo persistence. However, systemic toxicity in vivo was also observed. Figure generated with BioRender.

The authors also report the first use of the TcBuster transposon system in combination with BE to generate multiplexed CAR-NK cells providing a non-viral manufacturing pipeline requiring only a single electroporation (figure 1). These engineered NK cells exhibited significantly enhanced functionality in vitro and in vivo. However, the study also highlights and questions the importance of early and detailed safety investigation, as the authors noted toxicity in the mice treated with multiplex-edited and IL-15-expressing CAR-NK cells.

Leveraging the authors’ prior work,⁷ the authors performed single-gene knockouts (KOs) by targeting splice sites of several NK cell specific immunomodulatory genes, including AHR (aryl hydrocarbon receptor) and CISH (cytokine-inducible SH2-containing protein), a known negative regulator of IL-15 signaling. They also targeted the inhibitory receptor KLRG1 (Killer cell Lectin-like Receptor G1) and immune checkpoints TIGIT (T cell immunoreceptor with immunoglobulin and ITIM domain, TIGIT) and PDCD1 (programmed cell death protein 1,PD-1). TIGIT and PD-1 are classical immune checkpoints and KOs in T cells have been described to improve resistance to tumor-mediated immunosuppression. Similarly, TIGIT-KO or PD-1 KO in NK cells has been proposed to improve (CAR) NK cell function in certain scenarios. However, their detailed functional role in NK cells is less understood. Up to 100% knockout was achieved using the eighth generation adenine base editors, ABE8e, leading to improved in vitro functionality of all edits, except in the case of KLRG1 no improved functionality was shown. In the multiplex BE experiments, the authors used up to six single guide RNA (sgRNAs) simultaneously, without significant loss of editing efficiency, supporting BE as the new standard for precise and scalable NK cell gene editing. This is of particular interest, as NK cells, unlike T cells, have traditionally been considered as hard-to-transfect. The presented protocol for electroporation of genome engineering tools is comparable with others, providing further evidence of suitability of non-viral gene-editing for scalable NK cell engineering.^{4 5} Using a Raji tumor in vitro co-culture model, the authors identified the most effective multiplex combination to enhance intrinsic NK cell killing. Interestingly, the triple knockout of TIGIT, PDCD1, and CISH (TPCko) resulted in the strongest in vitro functional improvement. As a preliminary assessment of genotoxicity, the authors analyzed editing at the top 8–10 computationally predicted off-target sites using rhAmpSeq. They identified one site with significant, although low-frequency, off-target activity for the gRNA targeting PDCD1. This site was located in a non-coding region with no known biological function. Additionally, they confirmed negligible rates of translocations between on-target sites. While further off-target analysis is warranted, current tools to nominate off-targets of base editors are limited and typical assays, like GUIDE-seq, CIRCLE-Seq or CAST-Seq, rely on the occurrence of nuclease-induced DSBs that are reduced in base editors over conventional programmable nucleases.⁸

The authors then combined TPCko with IL-15-expressing CARs delivered by the TcBuster system, yielding a highly efficient, fully non-viral engineering strategy for multiplexed CAR-NK cells. These IL-15-equipped cells significantly outperformed the other treatment groups in vitro. The first xenograft mouse model confirmed IL-15 mediated support of CAR-TPCko NK cells in vivo, but the occurrence of marked toxicity by the cell therapy prevented statistical improvements in overall survival. Importantly, only the CAR15 with TPCko-treated group showed long-term persistence of NK cells in the blood, bone marrow, and spleen. In contrast, the CAR15-only or TPCko-only groups did not persist in the model, highlighting a potential synergistic benefit between edits and IL-15 for longevity and therapeutic efficacy of NK cells. However, the authors also report systemic toxicity. In the CAR15/TPCko group 30-50% of the mice died due to suspected toxicity (weight loss >20%). The criterion for toxicity was determined as low tumor region of interest (ROI) signal and >20% weight loss or death at <75 days, but the mechanisms have not been further investigated. A second larger xenograft experiment with more mice demonstrated less toxicity and statistical evidence regarding improved overall survival of the CAR15/TPCko group over the CAR15 only treatment. Monitoring of the mice revealed a more rapid NK cell expansion in the CAR15/TPCko group shortly before toxicity occurred. The exact cause of toxicity could not be delineated, and further investigation is required before clinical translation. Taking both animal studies into account, the therapeutic benefit of the chosen modifications is not yet definite.

This could raise the question of whether the combination of the specific multiplex editing, combined with IL-15 expression triggered the NK cell-mediated toxicity. Other studies have reported that NK cells engineered to express IL-15 induced significant toxicity in glioblastoma xenograft models in mice, including marked weight loss and extensive NK cell infiltration-effects that were not observed with transgenic IL-21.⁹ IL-15 cytokine-armored CAR-NK cells, which secrete IL-15 to maintain anti-Acute Myeloid Leukemia (AML) activity, have been associated with adverse systemic effects in xenograft models in vivo.¹⁰ However, all current clinical trials using CD19-CAR expressing NK cells with IL-15 have not yet reported such adverse effects.³ Therefore, it remains uncertain whether the toxicity observed by Wang et al would translate to patients, as the study was conducted in an immunodeficient xenograft mouse model. In a more immunocompetent setting, the persistence of the modified NK cells may be reduced, potentially mitigating the observed toxicities. However, this hypothesis has yet to be tested and underscores the importance of employing more physiologically relevant humanized models during preclinical evaluation. However, the improved durability of IL-15-modified NK cells offers potential evidence for overcoming key limitations of NK cell therapies, because enhanced persistence has been linked to better long-term outcomes in a clinical trial.³

Taken together, the important work by Wang et al highlights the tremendous potential of non-viral, multiplex base editing as a next-generation strategy for engineering CAR-NK cells with enhanced intrinsic cytotoxicity. At the same time, it underscores the critical need for rigorous and comprehensive safety assessments of multiplex-edited cell products, particularly as the field advances toward increasingly sophisticated engineering approaches for allogeneic NK cell therapies in the coming years.